experience-app-mcp-generate
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes the Salesforce CLI (
sf) to perform project discovery, query org metadata, and deploy changes to the Salesforce environment. Automation scripts likeactivate_server.mjsandpreflight.mjsuseexecFileSyncto runsfcommands. Parameters such as org aliases and developer names are validated against strict regex patterns (e.g.,isValidDeveloperName) to prevent shell injection. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted local source files and remote org data to generate server configurations, representing an indirect injection surface. Ingestion points: The skill reads local ExperienceBundle JSON, Apex classes, and Flow metadata, as well as remote action descriptors from the Salesforce API via
retrieve_existing_tools.mjs. Boundary markers: The instructions require the agent to show diffs and obtain explicit user consent before writing files or deploying to the org. Capability inventory: The skill is capable of writing to the local filesystem (fs.writeFileSync) and modifying the Salesforce org's state through deployment and Tooling API writes. Sanitization: Scripts likemerge_tools.mjsandrender_eca.mjsimplement XML entity escaping (escapeXml) to ensure that dynamic content is safely interpolated into metadata templates.
Audit Metadata